The Walk Forward Analyzer (WFA) module is designed to enhance your back-testing arsenal by providing the ability to optimize your results through a process of walk-forward analysis. A walk-forward analysis is conducted on a specific Expert Advisor with a nominated Metatrader terminal. The analyzer utilizes the in-built Optimization feature of Metatrader to optimize the results against the in-sample set of data. Once the best result has been found (or manually chosen) across the defined parameter space, the analyzer will proceed to use the Strategy Tester to perform an out-of-sample test. The testing continues for the specified number of iterations - each iteration being one sequence of an in-sample optimization followed by the out-of-sample test. Upon completion of the test, the analyzer with present a report of the results.
Walk Forward Analysis (WFA) is a method of optimizing your trading system while ensuring that you do not over-optimize your settings for a particular date range. An over-optimized system may perform very well on a set of known ("in-sample") data, however when presented with new ("out-of-sample") data it hasn't seen before, it may perform very poorly. This phenomenon is known as "curve fitting" since the strategy has been tailored to work on a known set of data. Walk Forward Analysis is a tool to avoid this curve fitting. It tests the the robustness and profitability of your trading strategy when presented with new and unseen out-of-sample data.
Walk Forward Analysis works by firstly breaking the testing date range into a series of smaller periods. For each period, it performs an optimization on a period of in-sample test data, followed by a set of out-of-sample data. The best performing settings for the out-of-sample data then become a candidate for the next in a series of iterative tests, walking forward through the set of available historical data. If the strategy performs well throughout its out-of-sample tests, there is some confidence that it will continue to work in a live environment.
The diagram below shows the series of iterative tests that are performed during the analysis through the selected historical date range.
|
The Walk Forward Analyzer works by utilizing the existing functionality of the Metatrader 4 Strategy Tester feature. The WFA invokes the Metrader Stategy Tester across historical data in an iterative walk-forward fashion. During each iteration, it will use the standard 'Optimization' feature of MT4 that will search for the best settings for that particular period. After this optimization phase, the WFA will automatically select (or allow you to manually select) the best-of-breed strategy settings that will be used for the an out-of-sample test. It will then produce a report on the series of results that it got during each test. This report can be used to understand the optimal configuration of your strategy.
Combined with the advanced testing capabilities of the Advanced Tester, the Walk Forward Analyzer assists with finding optimal settings for a trading strategy and gives a deeper insight into the potential performance the strategy. This includes being able to see how various simulated market conditions such as slippage and spread affect your strategy.
|
The Walk Forward Analyzer (WFA) module is designed to enhance your back-testing arsenal by providing the ability to optimize your results through a process of Walk Forward Analysis. The walk-forward module screen is described below.
The following elements comprise the Walk-forward Analyzer:
1.Walk-forward Analysis tab - Click this tab to activate the Walk-Forward Analyzer 2.Main Panel - The Walk Forward Analysis settings control panel. 3.Status box - Gives the current status/progress of the analysis. 4.Advanced Settings - The advanced test settings to use during the walk-forward analysis.
Please refer to the guide on Performing an Analysis for more details on how to use the analyzer.
|
Further reading
Please consult the Metatrader Manual on Optimization for further information on their optimization feature.